We just closed the books on 2025, and it was a wild year for capital.

In March, SoftBank wrote a $30 billion check to OpenAI. By December, that wasn't even the most interesting deal of the year.

Here's the breakdown.

The Top 10 Fundraises

1. OpenAI: $40B  

The largest private investment ever.
SoftBank wrote a $30B check. ChatGPT currently processes more traffic than most search engines handled in 2020 and has 900M weekly active users, as per Backlinko.

2. Scale AI: $14.3B  

Meta bought 49% and hired the CEO.
This was an acquisition dressed as a funding round. Meta paid $14B to own the pipes that label training data for every foundation model.

3. Anthropic: $13B  

Claude Code launched in May, hit a $1 billion in run-rate revenue in only 6 months. That velocity justified the September round.

4. Databricks: $15.3B  

Second Meta mega-check. Pattern emerging: Meta is buying the infrastructure layer before anyone else locks it down.

5. Anduril: $2.5B  

The only non-AI company in the top 10. Palmer Luckey's defense tech bet. Autonomous drones, DoD contracts, and the clearest signal that defense budgets are flowing toward AI-native companies.

6. Cursor: $2.3B  

AI code editor valued at $29.3B. Founded 2022. Developers pay $20/month for AI to write functions. 

7. Polymarket: $2B  

The NYSE owner invested in Polymarket, the same platform that processed $3.2B in election bets during the 2024 cycle. Institutions suddenly want programmable prediction infrastructure.

8. Thinking Machines: $2B  

Largest seed ever. No product, no revenue, four months old. Mira Murati left OpenAI as CTO, and raised $2B in her first round. The OpenAI alumni network is worth more than most venture portfolios.

9. Mistral AI: $2B  

Europe's only entry in the list, backed by Nvidia and ASML. Open-source models competing with closed US alternatives. Semiconductor firms hedging against a world where three US companies control foundation models.

10. Safe Superintelligence: $2B

Ilya Sutskever, the former OpenAI chief scientist, left to build safe AI and raised $2B from Alphabet and Nvidia months after founding. Second OpenAI alum to raise $2B+ on reputation alone.

What This Tells Us

  1. Foundation models dominated: Five companies building core AI models raised ~$91B combined. This was a land grab for the base layer.

  2. Infrastructure mattered more than ideas: Meta alone deployed ~$25B into data, storage, and tooling. Whoever owns the pipes controls the flow.

  3. Talent became the product: Founders with proven track records raised historic rounds before shipping anything.

  4. Tools caught up to platforms: Developer tools and infrastructure companies are now valued on par with model builders because adoption lives there.

Three things we're watching in 2026

  1. AI tools proved their value in 2025: This year showed us what AI can do. In 2026, the focus shifts to who gets access, who can build on top, and who benefits from the upside.

  2. Web3 becomes the access layer for AI: As AI power concentrates, demand grows for open systems, such as shared ownership, transparent incentives, and global participation. Web3 provides the rails to make that possible.

  3. Private markets start to open up: AI-native ventures need faster paths to capital, ownership, and liquidity. Fractional access and programmable markets move from edge cases to core infrastructure.

2025 gave us plenty to be excited about. Here’s hoping 2026 brings even more wins, better ideas, and a little more upside for everyone building in this space. 

Wishing you a strong, curious, and rewarding year ahead.

Happy New Year

from the VentureVerse team.

Keep reading